Without feedback, PWM is like using a linear regulator (like the 317) ... of course with PWM, there's no power disipation in the semiconductor, and it's more efficient for battery life (the motor is not always on ...)
Once I tried to make an speed regulator for a walkman motor, with a variable output regulator ... You can get the thing working fine and adjusted ... but as I keep testing the speed went slow and slow. I adjust it and again the same.
Finally I realized that a fixed voltage isn't enough to keep RPM constant. Of course .. if you have an external pot, you move until you are pleased and you can update the adjust it in case you need. It would be nice some strobo to check that's adjusted ... like the turntables
Not sure but the common solution used in motors like mabuchi are a a variable voltage supply with some kind of correction using the sensed current. If current raises, that's because load had raisen. So more voltage is applied to the motor in order to keep the RPM constant. But motor manufacturers don't know the whole system (load, inercia ...) so this method isn't perfect. It's a big lay ... no real feedback ... but at least, current had been taken as a data. Hovewer, this babys had worked fine in our boomboxes ... cheap but still good.
I'll try to measure a mabuchi motor core voltage while changing the load .... That's an interesting homework !!!!
For those boomboxes with two speeds, I agree with Ramon. Take a motor without controller and put the controller in the main board.
Ok ... a motor with feedback is intended for high quality decks ..
still love the idea though ....